[Engineer[_2_]]

Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) With no pre-amp connected the hum and noise
just went up as the pot was tuned up. The penny dropped! The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

[Alex Pogossov]

"Engineer" wrote in message
...
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) With no pre-amp connected the hum and noise
just went up as the pot was tuned up. The penny dropped! The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

Just by the way: If a pot is set to midpoint, and source output impedance is
low, then output impedance of the wiper is R/4, not R/2.

But... obviously the smaller the pot value -- the less hum and less grid
current related noise will be there anyway.

[GRe]

"Alex Pogossov" wrote in message
...

"Engineer" wrote in message
...
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) With no pre-amp connected the hum and noise
just went up as the pot was tuned up. The penny dropped! The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

Just by the way: If a pot is set to midpoint, and source output impedance
is low, then output impedance of the wiper is R/4, not R/2.

Just by the way: R/4 at wiper midpoint only when it's a linear pot, not log.

[glenbadd]

Is there an input coupling capacitor? Using a lower value pot will increase the high pass rollover frequency.

[John L Stewart]

Originally Posted by GRe

"Alex Pogossov" wrote in message
...

"Engineer" wrote in message
...
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) With no pre-amp connected the hum and noise
just went up as the pot was tuned up. The penny dropped! The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

Just by the way: If a pot is set to midpoint, and source output impedance
is low, then output impedance of the wiper is R/4, not R/2.

Just by the way: R/4 at wiper midpoint only when it's a linear pot, not log.

Only if driven by a low Z source, which it probably will be.

[Patrick Turner]

On Sep 28, 11:40*am, Engineer wrote:
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) *It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) *With no pre-amp connected the hum and noise
just went up as the pot was tuned up. *The penny dropped! *The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. *But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

The noise you get with 1M in series between G1 and 0V is the amplified
sound of the resistance noise. There is a formula to calculate the
noise and it varies with square root of bandwidth, and with square
root of resistance value. Thus if you reduce the BW by factor = 1/4,
then noise is only halved, and if you reduce resistance by say 1/4,
noise is only halved.

So that if R is reduced by 1/20, noise is reduced 1/4.4. If BW is
reduced 1/100 say from 20kHz to 200Hz, noise is reduced by 1/10.

The high Z grid input terminated by 1M also picks up electrostatic
effects, higher harmonics from rectifiers especially.

Using input resistance = 10k is quite OK for a tube amp with CF, even
if it a weak 12AX7 with Ia only 0.7mA. If you get only 0.2mA od peak
Ia change, its 2Vpk at the output, and enough to deafen you in most
power amps which clip at 1Vrms input. Do the sums. Distortion will be
low, hence the rule, at least 600 ohms Rout for line level components,
and at least 10k Rin.

I've often used a 20k log pot or DACT attenuator switch and the sound
remains as good as it can be. The simplest form of preamp is a passive
type with source switch feeding a 20k log pot. If its fed with 600
ohms, Rout at -6dB, ie, 1/2 max Vo, then the terminating Rout from the
20k pot = 5k0, and this is 1/50 times less than if you had a 1M pot
set at -6dB, so noise is reduced x 1/7, or -17dB.
Noise from 5k0 resistance with 20kHz BW might be 4uV, and I'm
guessing, so if the power amp gain is 20x, then noise from R = 80uV.
Usually, noise from the power amp with input taken to 0V with short
lead **should** be 0.5mV if 8 ohm speakers rated for 88dB/W are to
be used at 3M distance. The extra 0.08mV from 5k0 won't be heard. But
with Rsource = 250k, noise at Vo would be 7 x 4uV x 20 560uV = 0.56mV,
so total noise = sq.rt ( 0.5squared + 0.56squared) = 0.75mV.
But you have to factor in hum pick up and burst noise and other crap
in V1.

I suggest you study carefully how to work all this out. You do call
yourself Engineer, so be one!!!!

When you place a series capacitor ahead of the Rg, you have a HPF
filter. So no matter how low the source resistance, at very low F the
amp input grid remains terminated by R between grid and ground as
determined by the fixed R if used, or by the pot setting from wiper to
0V.
Resistors make their noise right down to DC, so with a HPF there can
be slight rumble adding itself into a system. If the input is
terminated with a coil with low dcr, then LF noise is shunted as XL
reduces. But coils are real good at picking up hum, and they always
cost, so they are never used at power amp inputs unless someone has a
specially designed input transformer to allow a floating balanced
input, usually to be driven by a 600 ohm source, which means XL should
not be less than 600 ohms at 16Hz, which means
L = 6H, quite a lot.

Say you have a phono amp with capacitance input to allow base of gate
biasing to be more conveniently arranged.
Well, if you wish to use a moving coil cart so the amp has to have
high gain, then you only need a very small amount of LF noise to get
considerable "cone wobbles" at LF. So R needs to be very low, maybe
1k0 max, and C = 10uF, and this means a large sized cap unless perhaps
your'e tempted to use a bibolar electrolytic. Electrolytic coupling
caps were routinely used for driving base inputs of bjts in bygone
days before quiet enough opamps came along. MC carts have very low
source Z 20 ohms, which means noise over 20kHz is about 1/16 of say
5k Z of a MM cart.

The trouble with electronics is that you must always think of 2 or
more interactive factors which often interact non linearly, if not
oppositely to common sense. That's why designing and building stuff
takes so long to optimise, and bosses and accountants and customers
and wives hate all that; they expect immediate and exciting solutions,
at -20dB lower cost. Well, I say **** 'em; let em wait until we've got
it right.

Patrick Turner.

[John L Stewart]

Gotta get on the bike now........too much noise here!!

Probably get rained on today.

Cheers, John

[Engineer[_2_]]

On Sep 28, 5:06*pm, "GRe" wrote:
"Alex Pogossov" wrote in message

...











"Engineer" wrote in message
....
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) *It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) *With no pre-amp connected the hum and noise
just went up as the pot was tuned up. *The penny dropped! *The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. *But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

Just by the way: If a pot is set to midpoint, and source output impedance
is low, then output impedance of the wiper is R/4, not R/2.

Just by the way: R/4 at wiper midpoint only when it's a linear pot, not log.

Alex, right, of course! With a zero Z source, you get 25K in parallel
with 25K at mid rotation of a linear pot, a quarter of the original
50K... at some other place further around for the log pot that I
used. My fault, in too much of a hurry... Thevenin rules!
Cheers,
Roger

[GRe]

"John L Stewart" wrote in message
...

GRe;939604 Wrote:
"Alex Pogossov" wrote in message
...-

[...]

Just by the way: If a pot is set to midpoint, and source output
impedance is low, then output impedance of the wiper is R/4, not R/2.-

Just by the way: R/4 at wiper midpoint only when it's a linear pot, not
log.

Only if driven by a low Z source, which it probably will be.

You're in the repeater business?


--
John L Stewart

[GRe]

"Engineer" wrote in message
...
Just by the way: If a pot is set to midpoint, and source output
impedance
is low, then output impedance of the wiper is R/4, not R/2.

Just by the way: R/4 at wiper midpoint only when it's a linear pot, not
log.

Alex, right, of course! With a zero Z source, you get 25K in parallel
with 25K at mid rotation of a linear pot, a quarter of the original
50K... at some other place further around for the log pot that I
used. My fault, in too much of a hurry... Thevenin rules!
Cheers,
Roger

Depending on a pot's taper and quality, for a log pot
ccw tap to wiper midpoint is approximately R/10.
With a zero Z source and approximately Rx0.9 cw tap
to wiper midpoint you'd get around R/11 wiper Z.

[Patrick Turner]

On Sep 29, 11:00*pm, John L Stewart John.L.Stewart.
wrote:
Gotta get on the bike now........too much noise here!!

Probably get rained on today.

Cheers, John

--
John L Stewart

I have got to ride after too very wet deighs hear.

I spent tyme rewiring a truly frightfully made stero amp made by Earl
Weston. Looks very pretty, goes like ****. Just one little drop and
you get lotsa splinters and broken glass..

Plus more work rwirin' Quad-II Forties. Nuttin's sacred 'round 'ere !
Singing lessons are compulsory curiculum for all amps entering the
shed door.Gently encouarged with hot soldering iron, long nose pliers,
side currers and a very stern chior leader.

Probably one of THE WORST things being inflicted by Chinese indistries
have been tube sockets.

In Quad-II-Forty, they have white ceramic sockets meant for PCB
because the contacts are just extended pins meant to poke into drilled
holes for soldering to tracks.
The trouble is the tube pin grippers. These are made zinc or cadmium
plated metal, not sure what metal, stamped to a Y shape with top of
the Y made to engage the tube pins. But the metal has very little
springiness, so after pushing a tube is and out 10 times, the tube
becomes quite loose as the distance between the Y increases. The Y
grippers have had two cuts in the metal bottom leg so that when
inserted into the slots in the ceramic, the bottom Y leg is given
twist so the leg section below the cuts turns out to prevent the pin
coming back up out of the slot its in. But you only have to bend the
bottom section of the pin gripper side ways say 45 degrees, and many
will just break off because the metal is so KUNTY and FUCTATIOUS.

It has been said they "don't make thigs like they used to" and this is
very true of many things relating to tube amps, and especially to such
things made in China where the ruling elite who run Chinese industries
to provide western customers with Quad amps costing many thousands of
dollars while paying the chinese workers only an average of 64c a day,
and while using **** METALURGY in critical items like tube sockets.

The Chinese tube sockets fitted into hi-end Chinese amps or sockets
bought by wannabe makers like Earl Weston are often utter garbage, and
I don't know if the situation is improving because the Chinese hate
being told to do better about product reliability. They think they are
superior, but I know their **** stinks just like anyone elses.

The Chinese have a long way to go before making tube sockets which are
as good as the McMurdo and other weston made sockets which last well
against many plug ins/outs, and having tubes pushed over while in
sockets which is particularly bad for the Chinese sockets. Good tube
sockets allow grippers to move and follow tube sways, without the
metal deforming, bending, or yielding.

While gutting the Quads in preparation for re-wiring my way, a contact
broke so very easily on a tube socket. I was able to repair it by
using some tinned wire 0.5mm dia, in a loop over the Y grippers, then
flattened with pliers and soldered. The two wire ends are taken down
through the slots in ceramic as the slots are wide enough. The
grippers can be bent with an awl carefully used to make the gap
smaller to grip tube pins tighter, but after I did this the fix didn't
seem to last long, and the tube still felt like not much was holding
it in place.

Audiophiles change tubes like underpants. They are always trying to
prevent ****ty sound by changing tubes, and they do themselves no good
because sooner or later the grippers will become intermittent and
cause smoke and silence.

The S&S brought the Earl Weston amp to me for a fix. Fulla chinese
tube sockets which fail so easily. Owner had tried a big 5R4 because
some idiot ****wick@somewhere had said bass would improve. He'd also
tried 5AS4. But 5AR4 is the best. A heavy tube like 5R4 nearly can
fall out of the socket if the amp is turned upside down. It is in-
evitable that amps like this get moved around as audiophiles lend them
to mates and sell them to one another. With a big 5R4 with only 4 4
grippers left in the socket, there isn't much to hold the tube, and
they easly get bent over during moves as there is no cover screwed
down over tubes. The rectifier heat cycles also tend to fatigue the
metal.

The Earl Weston had KT66 with 43% UL taps on OPT, in class A, for 14
watts with Ea at 350V. But just 1/2 6SN7 input powering one OP grid
with a 1/2 6SL7 driving the other grid, with this single 1/2 6SL7 a
floating self balancing phase inverter circuit. So the other 1/2 6SL7
remained unused. NO GNFB applied, so Rout measured 10.6 ohms, noise&
distortion all too high, but who cares anyway after they spend $5,000.
The wooden chassis base using 18mm thick sides is very poor carpentry,
and stainless steel top-plate and bottom plate held to wood with only
4 screws, so one drop and the whole damn thing flies apart into
splinters. There was a signed note on the inside of the woodwork that
the amp was supplied as a replacement for another that had been dmaged
in transit, and that a warranty was not offered. So along with the
poor workmanship, no good will either. The world is fulla ppl making
lousy quality garbage because they won't put the effort in that real
quality demands, and its because more effort means less earnings.

I'm replacing all circuitry with equivalent of my 5050 schematic which
works so much better than Earl the Charlatan could ever manage.

What gives good bass is adequate PSU capacitance and low Rout, ie,
good damping factor, by means of class A triodes OR else adequate GNFB
for the UL connection.

So folks, there is a huge amount of difference in tube sockets, and
there are many gullible people who would not recognise a decent tube
socket even if it bit them on their arse !. The problem of poor
sockets isn't monopolised by the Chinese though. I completely re-wired
an ARC VT100 some years ago, and its tube sockets were so bad tubes
would creep out of sockets. ARC had 8 horizontally mounted tubes in a
close box, so covers had to be unscrewed to push tubes in. I soldered
0.8mm wire to bases of 6550 and lashed the tubes in by tying the wire
down around the socket stand-offs, so no more problems with tube
creep.
The ARC also gave lots of Smoke&Silence, and the complete rewire
changed all that.

Noys hear isn't two bad though, cood be wirse, three or four bad.
But when I ryde, I nough I am pe-sewd by Uncle Fate and peneenless
customers close behynde.

Patrick Turner.

[John L Stewart]

Originally Posted by Patrick Turner

On Sep 28, 11:40*am, Engineer wrote:
Hi, Vcuumlanders,
Perhaps this is just a minor point, or even common knowledge, but it
has merit...
I had put a pre-set 1 meg log gain pot at the front end of my recently
built monobloc (the Williamson near-clone that some here may
recall...) *It worked, of course, but I was aware that near the mid-
point setting the amp had more hum and noise than at either end (with
a pre-amp connected.) *With no pre-amp connected the hum and noise
just went up as the pot was tuned up. *The penny dropped! *The grid of
the first audio tube was picking up "under-chassis" hum and noise when
the first stage grid had a high impedance to ground, i.e. 1/2 meg, or
so. I had not specifically shielded any under chassis AC wiring and
there was a 60 Hz power transformer under the chassis, too, for B+
boost.
The solution: use a 47 K log pot for the preset input gain. Now the
first grid never sees more than 23 K to ground - very quiet. *But now
the amplifier now has 47K input impedance. This is OK if the pre-amp
has a cathode follower output, i.e. has a Thevenin impedance of, say,
400 ohms (it has - it's a Heathkit WA-P2), or if the pre-amp is solid
state - 47K is easy to drive.
I promptly changed my two other monoblocs, a home-brew and a Pilotone
commercial unit, to the same pre-set pot input value with some noise
improvement - less than the first one, but the idea is a sound
one.
Cheers,
Roger

The noise you get with 1M in series between G1 and 0V is the amplified
sound of the resistance noise. There is a formula to calculate the
noise and it varies with square root of bandwidth, and with square
root of resistance value. Thus if you reduce the BW by factor = 1/4,
then noise is only halved, and if you reduce resistance by say 1/4,
noise is only halved.

So that if R is reduced by 1/20, noise is reduced 1/4.4. If BW is
reduced 1/100 say from 20kHz to 200Hz, noise is reduced by 1/10.

The high Z grid input terminated by 1M also picks up electrostatic
effects, higher harmonics from rectifiers especially.

Using input resistance = 10k is quite OK for a tube amp with CF, even
if it a weak 12AX7 with Ia only 0.7mA. If you get only 0.2mA od peak
Ia change, its 2Vpk at the output, and enough to deafen you in most
power amps which clip at 1Vrms input. Do the sums. Distortion will be
low, hence the rule, at least 600 ohms Rout for line level components,
and at least 10k Rin.

I've often used a 20k log pot or DACT attenuator switch and the sound
remains as good as it can be. The simplest form of preamp is a passive
type with source switch feeding a 20k log pot. If its fed with 600
ohms, Rout at -6dB, ie, 1/2 max Vo, then the terminating Rout from the
20k pot = 5k0, and this is 1/50 times less than if you had a 1M pot
set at -6dB, so noise is reduced x 1/7, or -17dB.
Noise from 5k0 resistance with 20kHz BW might be 4uV, and I'm
guessing, so if the power amp gain is 20x, then noise from R = 80uV.
Usually, noise from the power amp with input taken to 0V with short
lead **should** be 0.5mV if 8 ohm speakers rated for 88dB/W are to
be used at 3M distance. The extra 0.08mV from 5k0 won't be heard. But
with Rsource = 250k, noise at Vo would be 7 x 4uV x 20 560uV = 0.56mV,
so total noise = sq.rt ( 0.5squared + 0.56squared) = 0.75mV.
But you have to factor in hum pick up and burst noise and other crap
in V1.

I suggest you study carefully how to work all this out. You do call
yourself Engineer, so be one!!!!

When you place a series capacitor ahead of the Rg, you have a HPF
filter. So no matter how low the source resistance, at very low F the
amp input grid remains terminated by R between grid and ground as
determined by the fixed R if used, or by the pot setting from wiper to
0V.
Resistors make their noise right down to DC, so with a HPF there can
be slight rumble adding itself into a system. If the input is
terminated with a coil with low dcr, then LF noise is shunted as XL
reduces. But coils are real good at picking up hum, and they always
cost, so they are never used at power amp inputs unless someone has a
specially designed input transformer to allow a floating balanced
input, usually to be driven by a 600 ohm source, which means XL should
not be less than 600 ohms at 16Hz, which means
L = 6H, quite a lot.

Say you have a phono amp with capacitance input to allow base of gate
biasing to be more conveniently arranged.
Well, if you wish to use a moving coil cart so the amp has to have
high gain, then you only need a very small amount of LF noise to get
considerable "cone wobbles" at LF. So R needs to be very low, maybe
1k0 max, and C = 10uF, and this means a large sized cap unless perhaps
your'e tempted to use a bibolar electrolytic. Electrolytic coupling
caps were routinely used for driving base inputs of bjts in bygone
days before quiet enough opamps came along. MC carts have very low
source Z 20 ohms, which means noise over 20kHz is about 1/16 of say
5k Z of a MM cart.

The trouble with electronics is that you must always think of 2 or
more interactive factors which often interact non linearly, if not
oppositely to common sense. That's why designing and building stuff
takes so long to optimise, and bosses and accountants and customers
and wives hate all that; they expect immediate and exciting solutions,
at -20dB lower cost. Well, I say **** 'em; let em wait until we've got
it right.

Patrick Turner.

The noise formula referenced predicts only the noise voltage or current resulting from an ideal resistor. In a real resister noise would be more.

Again, in a real cct more random noise is generated by the input characteristic of whatever gain device is used, be it SS or vacuum. For vacuum it turns out triodes are best while running lots of plate current. The cascode arrangement is quite good. I've used that in the past for input from reluctance pickups, where only a few millivolts signal are available. I used RF triodes such as the 6BK7/6BQ7/6BZ7 family with much success. 6DJ8 & 6AQ8 should work very well also.

But for the input of a power amp a little more noise is not a problem. Hum is. Here I would use an EF86 or Z729 if it was critical. But I've found rather ordinary things like the 6AU6 do well in most applications.

Others may differ.

But I'm sure that all this is well known here!!

Biked another 50 km today. Looking for a 600 mile Septmber by this time Friday afternoon. Weather don't look good. Cold & wet predicted.

Cheers, John

[Patrick Turner]

John mentioned...
The noise formula referenced predicts only the noise voltage or current
resulting from an ideal resistor. In a real resister noise would be
more.

In my tests of tubes and noise I have found the resistance value
dominates the noise production and that the formulas are about right.

My tests for noise in tubes and any added grid resistance were done
using a sample triode with grounded or ungrounded Rg, then amplifying
the anode signal using a quiet amp with gain of at least 1,000 x or
+60dB.
If the tube under test has gain = 16, then you know the noise seen at
the output of the 60dB amp is about 16,000 times higher than any grid
signal. So if grid noise = 1uV, there will be 16mV at the 60dB amp. If
a resistance is added then a calculation can be made to find its
noise.
N total = sq root ( NRg squared + other N squared ). Perhaps cheap old
carbon composition resistors are more noisy than metal film types.


Again, in a real cct more random noise is generated by the input
characteristic of whatever gain device is used, be it SS or vacuum. For
vacuum it turns out triodes are best while running lots of plate
current. The cascode arrangement is quite good. I've used that in the
past for input from reluctance pickups, where only a few millivolts
signal are available. I used RF triodes such as the 6BK7/6BQ7/6BZ7
family with much success. 6DJ8 & 6AQ8 should work very well also.

The noise in tubes varies much more than it varies for the same value
of resistance.

A good 1/2 12AX7 sample will have 1uV of grid noise, and if you have
dc on the heater and amplify the noise, its like a sputtery barbecue,
with considerable LF burst noise, ie, the tube noise has more LF
content than HF content. This prohibits 12AX7 being good enough for MC
carts. Gm = approx 1.2mA/V, and noise reduces with square root of
increase in Gm, so that if you had 5 x 12AX7 with all 10 triodes in
parallel, Gm = 12mA/V,
10 times the single 1/2 triode. One would expect noise to be 0.316
times less, but damned if I've ever see such a reduction. With a 6DJ8
with Ia = 5mA, Gm = approx 6.6mA/V, and both halves give Gm = 13mA/V,
and yet noise rarely measures any better than the single 1/2 12AX7.
Using a cascode does NOT improve anything
because the grid noise is still amplifed by the cascode gain. What the
cascode does do is reduced the triode Miller effect because the bottom
tube has very low gain drinving the top tubes whose gain = anode RL /
tube gain. The total cascode gain works out at gm x RLa, so for
cascoded 6DJ8, with Gm = 6mA/V, and RLa = 20k,
expect gain = 120, and if the top tube's gain = 18, then bottom tube's
gain = 6.6, as the cathode input resitance = 1.48k, and gain for load
= 1.48k = 6.6. ( Ra = approx 5k ).

Some ppl have used MC input to the cathode of a grounded grid tube.
This can be tricky and biasing the grid negatively will reduce Ek to
0.0V and allow the cart to be between cathode and 0V without any DC
flow, or very little. Noise rules still apply.

But the quietest cascode comes when a high Gm J-fet is used to drive a
triode or another J-fet.
Typical noise figures for 2SK147 driving say 1/2 6dJ8 are -20dB lower
than tube grid inputs. If RLa = say 16k,
and Ia = 5mA, gain = Gm x RL = 0.04A/V x 16k0 = 640. Usually an
unbypassed Rsource is used to reduce this gai to about 200. The j-fet
noise does not have high LF sputter noise of the tube, so when used in
an MC amp
or mircrophone amp the fet is a very effective device to minimise
noise.


But for the input of a power amp a little more noise is not a problem.

Indeed this is the case, and most good tubed power amps exhibit quite
adequate input noise figures.
If amp gain = 20, and input noise = 5uV from a not so wonderful input
tube, then output noise = 100uV, plus whatever else is generated
within the PO, and yes, its often mainly hum. If there is dc to all
heaters, and the PSU is remote, then the hum will be largle absent and
noise is very low. I've made 55Watt amps using a pair of 845, and had
noise 0.25mV with input grounded. 55W into 5 ohms = 16.5Vrms, and if
the noise does not increase for mas PO, the unweigted SNR -100dB.


Hum is. Here I would use an EF86 or Z729 if it was critical. But I've
found rather ordinary things like the 6AU6 do well in most
applications.

Apart from good B+ rail filters, DC on the heaters of all input/driver
tubes in power amps is the best way to reduce hum noise.

Others may differ.

But I'm sure that all this is well known here!!

Biked another 50 km today. Looking for a 600 mile Septmber by this time
Friday afternoon. Weather don't look good. Cold & wet predicted.

It rained and blew hard yesterday here and the 60km ride was cold and
miserable. My ankle has decided to be a real pest - it was badly
smashed up in a motorcycle accident 45 years ago, and maybe it tips me
off the bike. I might have to buy a kayak to get exercise if I can't
use the legs.
But in a month it will be quite warm here.

We don't export our spring sunshine.

But I'd gladly export a magpie which attacked me yesterday. Between
August and October its the breeding season, and they become very
teritorial, and its usual to get attacked by 2 or 3 during a ride, so
over the years I've probaby been attacked hundreds of times. But
yesterday's bird was the worst because it pecked me 20mm from my eye,
and would have got my eye were it not for wrap around sun-glasses. A
child here was recently blinded this way. If I manage to grab this
bird in my next ride when he attacks, I'll gladly thrash it to death
quickly, as its genes are way too agressive. Magpies do very well in
suburbia and city landscapes where they benefit from so much junk food
scraps littering the land scape. The road kill of animals on country
roads is a smorgasboard of food for maggies in spring, so they like to
nest in trees near roads, and from there they terrorise passing
cyclists. Putting a few cable ties in the helmet helps keep most away
but some come in from the side, then upwards to get eyes. I'll have to
apply to the Pentagon to get one or two spare drone aircraft to
accompany me on rides to fire warning missiles at the maggies. But I
reckon the maggies would attack the drones and win. Maybe there's a
real good market to export magpies to Ghan, where they could be used
against the Taliban - have 'em on their knees in the dust begging for
a peace deal and handing in all their AK47s and IEDs.

Patrick Turner.



Cheers, John

--
John L Stewart- Hide quoted text -

- Show quoted text -